Best female chess player

Polgar, January 2004: 2746
Gaprindashvili, June 1988: 2614
Menchik, May 1929: 2535

Obviously noone comes close to Judit Polgar

@tpr said in #9:

The statistician Jeff Sonas took on the giant task of calculating elo ratings backwards in time based on all published game results.
While I appreciate the effort, I still believe it only makes sense to compare only ratings at a specific moment in time or between times which are close (within few years).

Even at the same time the ratings should not be taken as religiously as people tend to take them. Everyone who played in some international tournaments can tell you that e.g. "1800 FIDE" can mean very different things in different countries. In competitions I often see players whose FIDE ratings are similar but their national ratings are 200 points apart or the other way around.

In August 2000 list, 15 players have rating 2700 or higher. In August 2025 it's 30 players (and people keep talking about "rating deflation problem" and need to solve it). Is rating of 2700 from 2000 supposed to be an equivalent of 2700 from 2025? Or is being ranked as No 15 an equivalent? It clearly cannot be both. People love having one magic number that can be used to compare easily what is "better" and what is "worse" but life rarely works like that.

"Is rating of 2700 from 2000 supposed to be an equivalent of 2700 from 2025?" * Yes. It means having 50% expected score against 2700 from 2012, or against 2700 from 2008 and 2700 from 2016.

"Or is being ranked as No 15 an equivalent?" No. Clearly not. Ranked No 15 in 1800 is clearly weaker than ranked No 15 in 2025.

"People love having one magic number that can be used to compare" * A number is objective. Preferring Morphy over Lasker, or Tal over Karpov is emotional, not rational.

Judit has merit of competing in par with top male's. Which is something that has not been repeated. That alone makes her clear #1.

@tpr said in #14:

... It means having 50% expected score against 2700 from 2012, or against 2700 from 2008 and 2700 from 2016. ...
Where is it verified that all these things are reliable indications of what would happen with players from nearly a century ago?

"Where is it verified that all these things are reliable indications of what would happen with players from nearly a century ago?"

• By the elo calculation Sonas conducted backwards in time based on all published chess game results.
  From now to a year ago, to 2 years ago, to 3 years ago... to 100 years ago.

@tpr said in #14:

A number is objective.
But often very misleading. And you often get very different results depending on which number you choose.

With CPUs, people used to take their frequency as the "magic number" which never worked reliably and it became very misleading even before multicore CPUs hit the mainstream. Now they tend to believe in benchmarks but, alas, different benchmarks often provide different results.

The same type of people used to compare car engines by their cylinder volume with similar results. In chess we have ratings - and it's still the same story.

@tpr said in #14:

... A number is objective. ...
Are beliefs about number-dependability necessarily objective?
@tpr said in #17:
... By the elo calculation Sonas conducted backwards in time based on all published chess game results.
From now to a year ago, to 2 years ago, to 3 years ago... to 100 years ago.
And accuracy measurements are found where?

@tpr said in #14:

"Is rating of 2700 from 2000 supposed to be an equivalent of 2700 from 2025?" * Yes. It means having 50% expected score against 2700 from 2012, or against 2700 from 2008 and 2700 from 2016.
No, it does not mean that at all. 2700 in 2000 list means 50% expected mean result against the pool it was calculated from, i.e. the 2000 pool. It says nothing about expected mean result against 2700 from the 2025 list calculated from results of a completely different 2025 player pool.

In August 1985, Garri Kasparov's rating was 2831. In August 2025, Magnus Carlsen's is 2839. Does it mean they would be equal if I were somehow able to let them play against each other? I don't think so.